10

I am planning on doing a script transformation utility (for extended diagnostic information) using Boost.Spirit 2.

While there is support for line information etc. for parsing errors, how i can store line numbers for successfully parsed expressions with Qi?

Georg Fritzsche
  • 97,545
  • 26
  • 194
  • 236
  • Hey gf I have a question open related to spirit http://stackoverflow.com/questions/2183067/boostspirit-grammar-for-unsorted-schema perhaps you know some way of doing this. – Hassan Syed Feb 02 '10 at 10:47

3 Answers3

4

As per the mailing list, Spirit.Classic positional iterators can also be used with Spirit 2.
There is also an article on an iter_pos-parser on the Spirit-blog.

I will update when i had time to test.

Georg Fritzsche
  • 97,545
  • 26
  • 194
  • 236
  • Quoting Joel Guzman from that article: _`..., if you don’t want to do any “Metacthulhu”, then this is the wrong article for you`_ / ... _`this is an advanced topic as people don’t normally want to extend Spirit and write their own components`_. – sehe May 19 '12 at 08:52
2

I realize I noticed the question late, but let me add this anyway. I wrote a sample of an INI file parser in another answer:

This uses a 'simple' semantic action with line_pos_iterator.

Here is the fruit of my labour: https://gist.github.com/1425972

  • When POSITIONINFO == 0
    • input is streaming
    • output is raw strings (well, map<string, map<string, string> > for the sections)

  • When POSITIONINFO == 1

    • input is buffered

    • output is textnode_t:

      struct textnode_t {
    int sline, eline, scol, ecol;
    string_t text;
};

      This means that the resulting map<textnode_t, map<textnode_t, textnode_t> > is able to report exactly what (line,col) start and end points mark the individual text nodes.

Here is a reduced demo. For full description and extensive test cases see the original anser or the code at github

Demo Input

[Cat1]
name1=100 #skipped

name2=200 \#not \\skipped
name3=   dhfj dhjgfd

Demo Output (POSITIONINFO == 0)

Parse success!
[Cat1]
name1 = 100 
name2 = 200 \#not \\skipped
name3 = dhfj dhjgfd

Demo Output (POSITIONINFO == 1)

Parse success!
[[L:1,C2 .. L1,C6:Cat1]]
[L:2,C2 .. L2,C7:name1] = [L:2,C8 .. L2,C12:100 ]
[L:6,C2 .. L6,C7:name2] = [L:6,C8 .. L6,C27:200 \#not \\skipped]
[L:7,C2 .. L7,C7:name3] = [L:7,C11 .. L7,C22:dhfj dhjgfd]
Community
  • 1
  • 1
sehe
  • 374,641
  • 47
  • 450
  • 633
0

Here is another useful article that explains how to use the exception that the phrase_parse function throws.

The article describes how to get error messages like this:

Exception: parse error at file STDIN line 1 column 10
'123,42.0,a,1.4'
          ^- here
Frank
  • 64,140
  • 93
  • 237
  • 324
  • You're bit late to the party. Also, your suggestion doesn't answer the question as it deals only with exceptions (specificly, `expectation_failure`). I posted **[an example](https://gist.github.com/1425972)** of how to use Spirit Qi to parse into nodes which store the source location _as well as_ the parsed value in **[this answer](http://stackoverflow.com/a/8365427/85371)** with extensive tests/samples. – sehe May 19 '12 at 08:49